agents.bayesianopponentmodel
Class PerfectBayesianOpponentModelScalable

java.lang.Object
  agents.bayesianopponentmodel.OpponentModel
      agents.bayesianopponentmodel.PerfectBayesianOpponentModelScalable

public class PerfectBayesianOpponentModelScalable
extends OpponentModel

Version of the standard Scalable Bayesian Model which uses the opponent's utilityspace to calculate the real utility of the opponent's bid. This is equivalent to having complete knowledge about the opponent's decision function. KNOWN BUGS: (similar to original BayesianOpponentModelScalable) 1. Opponent model does not take the opponent's strategy into account, in contrast to the original paper which depicts an assumption about the opponent'strategy which adapts over time. 2. The opponent model becomes invalid after a while as NaN occurs in some hypotheses, corrupting the overall estimation.

Author:

Mark Hendrikx

Field Summary

Fields inherited from class agents.bayesianopponentmodel.OpponentModel

fBiddingHistory, fDomain

Constructor Summary

PerfectBayesianOpponentModelScalable(UtilitySpace pUtilitySpace)

Method Summary

double	getExpectedEvaluationValue(Bid pBid, int pIssueNumber)
double	getExpectedUtility(Bid pBid) Plan: cache the results for pBid in a Hash table.
double	getExpectedWeight(int pIssueNumber)
double	getNormalizedWeight(Issue i, int startingNumber)
boolean	haveSeenBefore(Bid pBid)
void	setOpponentUtilitySpace(UtilitySpace opponentUtilitySpace)
void	updateBeliefs(Bid pBid)
void	updateEvaluationFns(double opponentUtility)
void	updateWeights(double opponentUtility)

Methods inherited from class agents.bayesianopponentmodel.OpponentModel

findMinMaxUtility, getDomain, getNormalizedUtility, isCrashed

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

PerfectBayesianOpponentModelScalable

public PerfectBayesianOpponentModelScalable(UtilitySpace pUtilitySpace)

Method Detail

getExpectedEvaluationValue

public double getExpectedEvaluationValue(Bid pBid,
                                         int pIssueNumber)
                                  throws java.lang.Exception

Throws:

java.lang.Exception

getExpectedWeight

public double getExpectedWeight(int pIssueNumber)

Overrides:

getExpectedWeight in class OpponentModel

updateWeights

public void updateWeights(double opponentUtility)
                   throws java.lang.Exception

Throws:

java.lang.Exception

updateEvaluationFns

public void updateEvaluationFns(double opponentUtility)
                         throws java.lang.Exception

Throws:

java.lang.Exception

haveSeenBefore

public boolean haveSeenBefore(Bid pBid)

Overrides:

haveSeenBefore in class OpponentModel

updateBeliefs

public void updateBeliefs(Bid pBid)
                   throws java.lang.Exception

Overrides:

updateBeliefs in class OpponentModel

Throws:

java.lang.Exception

getExpectedUtility

public double getExpectedUtility(Bid pBid)
                          throws java.lang.Exception

Plan: cache the results for pBid in a Hash table. empty the hash table whenever updateWeights or updateEvaluationFns is called.

Overrides:

getExpectedUtility in class OpponentModel

Parameters:

pBid -

Returns:

weeighted utility where weights represent likelihood of each hypothesis

Throws:

java.lang.Exception

getNormalizedWeight

public double getNormalizedWeight(Issue i,
                                  int startingNumber)

setOpponentUtilitySpace

public void setOpponentUtilitySpace(UtilitySpace opponentUtilitySpace)